需求模板一致性智能化检查  

INTELLIGENT CHECKING OF REQUIREMENT TEMPLATE CONFORMANCE

作　　者：张龙阳 杨卫东[1] Zhang Longyang;Yang Weidong(School of Computer Science,Fudan University,Shanghai 201203,China)

机构地区：[1]复旦大学计算机科学与技术系,上海201203

出　　处：《计算机应用与软件》2023年第9期290-295,302,共7页Computer Applications and Software

摘　　要：需求模板的一致性检查对于需求工程的自动化分析十分重要。现有方法往往需要通过术语表对需求语句的各个组件进行定位,或是通过单词词性组合来判定一致性信息,其在术语表缺失的情况下和在句式较长的需求语句上表现不佳。针对这种情况,提出基于依存关系的模板一致性检查方法,分析需求模板的依存关系来定义通用的依存语法模式,进而通过模式匹配的方法来判定需求语句的一致性信息,同时根据需求模板的特性来对词性标注和依存分析任务带来的错误进行校正,提高方法的准确率。经过实验验证,依存分析的方法相较于现有方法有了明显提升。The consistency check of requirement template is very important for the automatic analysis of requirement engineering.The existing methods often need to locate each component of requirement statement through glossary,or determine the consistency information by part of speech combination of words,which does not perform well in the case of missing glossary or in the requirement statements with long sentence patterns.In view of this,we propose a template consistency checking method based on dependency relationship.The dependency relationship of requirement template was analyzed to define the general dependency syntax pattern.The consistency information of requirement statement was determined by pattern matching method.At the same time,the errors caused by part of speech tagging and dependency analysis tasks were corrected according to the characteristics of requirement template,so as to improve the accuracy of this method.The experimental results show that the dependency analysis method is much better than the existing methods.

关 键 词：需求模板 一致性 自动化分析 依存关系 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]